2007-10-24 09:12:54

by Andi Kleen

Subject: [PATCH] Remove bogus default y for DMAR and NET_DMA

Andrew, please queue for next merge. Thanks.

----

No reason I can think of of making them default y
Most people don't have the hardware and with default y they
just pollute lots of configs during make oldconfig.

Signed-off-by: Andi Kleen <[email protected]>

Index: linux-2.6.24-rc1-hack/arch/x86_64/Kconfig
===================================================================
--- linux-2.6.24-rc1-hack.orig/arch/x86_64/Kconfig
+++ linux-2.6.24-rc1-hack/arch/x86_64/Kconfig
@@ -753,7 +753,6 @@ config PCI_DOMAINS
config DMAR
bool "Support for DMA Remapping Devices (EXPERIMENTAL)"
depends on PCI_MSI && ACPI && EXPERIMENTAL
- default y
help
DMA remapping (DMAR) devices support enables independent address
translations for Direct Memory Access (DMA) from devices.
Index: linux-2.6.24-rc1-hack/drivers/dma/Kconfig
===================================================================
--- linux-2.6.24-rc1-hack.orig/drivers/dma/Kconfig
+++ linux-2.6.24-rc1-hack/drivers/dma/Kconfig
@@ -43,7 +43,6 @@ comment "DMA Clients"
config NET_DMA
bool "Network: TCP receive copy offload"
depends on DMA_ENGINE && NET
- default y
help
This enables the use of DMA engines in the network stack to
offload receive copy-to-user operations, freeing CPU cycles.

2007-10-24 09:24:35

by Jeff Garzik

Subject: Re: [PATCH] Remove bogus default y for DMAR and NET_DMA

Andi Kleen wrote:
> Index: linux-2.6.24-rc1-hack/arch/x86_64/Kconfig
> ===================================================================
> --- linux-2.6.24-rc1-hack.orig/arch/x86_64/Kconfig
> +++ linux-2.6.24-rc1-hack/arch/x86_64/Kconfig
> @@ -753,7 +753,6 @@ config PCI_DOMAINS
> config DMAR
> bool "Support for DMA Remapping Devices (EXPERIMENTAL)"
> depends on PCI_MSI && ACPI && EXPERIMENTAL
> - default y
> help
> DMA remapping (DMAR) devices support enables independent address
> translations for Direct Memory Access (DMA) from devices.

ACK


> Index: linux-2.6.24-rc1-hack/drivers/dma/Kconfig
> ===================================================================
> --- linux-2.6.24-rc1-hack.orig/drivers/dma/Kconfig
> +++ linux-2.6.24-rc1-hack/drivers/dma/Kconfig
> @@ -43,7 +43,6 @@ comment "DMA Clients"
> config NET_DMA
> bool "Network: TCP receive copy offload"
> depends on DMA_ENGINE && NET
> - default y
> help
> This enables the use of DMA engines in the network stack to
> offload receive copy-to-user operations, freeing CPU cycles.

ACK -- but its arguable that given the current code, its worth setting
this if DMA_ENGINE is also set.

2007-10-24 15:30:44

by Shannon Nelson

Subject: RE: [PATCH] Remove bogus default y for DMAR and NET_DMA

>From: Jeff Garzik [mailto:[email protected]]
>
>Andi Kleen wrote:
>> Index: linux-2.6.24-rc1-hack/arch/x86_64/Kconfig
>> ===================================================================
>> --- linux-2.6.24-rc1-hack.orig/arch/x86_64/Kconfig
>> +++ linux-2.6.24-rc1-hack/arch/x86_64/Kconfig
>> @@ -753,7 +753,6 @@ config PCI_DOMAINS
>> config DMAR
>> bool "Support for DMA Remapping Devices (EXPERIMENTAL)"
>> depends on PCI_MSI && ACPI && EXPERIMENTAL
>> - default y
>> help
>> DMA remapping (DMAR) devices support enables
>independent address
>> translations for Direct Memory Access (DMA) from devices.
>
>ACK
>
>
>> Index: linux-2.6.24-rc1-hack/drivers/dma/Kconfig
>> ===================================================================
>> --- linux-2.6.24-rc1-hack.orig/drivers/dma/Kconfig
>> +++ linux-2.6.24-rc1-hack/drivers/dma/Kconfig
>> @@ -43,7 +43,6 @@ comment "DMA Clients"
>> config NET_DMA
>> bool "Network: TCP receive copy offload"
>> depends on DMA_ENGINE && NET
>> - default y
>> help
>> This enables the use of DMA engines in the network stack to
>> offload receive copy-to-user operations, freeing CPU cycles.
>
>ACK -- but its arguable that given the current code, its worth setting
>this if DMA_ENGINE is also set.
>

The intent is to give us a DMA offload for NIC-to-user operations. At
the moment the ioatdma device is the only gizmo supporting this through
NET_DMA, and is the only DMA_ENGINE supported on x86. As more devices
are made available through DMA_ENGINE on platforms that don't have
ioatdma, removing the "default y" makes sense. I agree that this is
ACK-able, but under the "it should just work" philosophy, and with the
Intel servers coming out with the hardware support, it may want to wait
until there are more DMA_ENGINE choices.

In any case, I'll ACK this and let the maintainers decide on the timing.

sln